Final-year diploma student based in Solapur, India. I build AI-driven tools and systems that solve real problems — not proof-of-concepts that sit in a repo and collect dust. My approach: engage with the problem directly, break things, iterate, and ship something that works.

I use AI agents as part of my actual development workflow — orchestrating task pipelines, automating repetitive engineering decisions, and treating LLMs as collaborators rather than tools you prompt once and walk away from.

Obsidian Graph Intelligence — An Obsidian plugin that treats your knowledge vault as a graph. Detects orphan notes, discovers topic clusters, and suggests missing links using offline Transformers.js embeddings. Ships with an adaptive learning system that refines recommendations from user feedback, a batch "Fix My Vault" repair engine, and optional LLM integration via Ollama, OpenAI, and Anthropic. 15+ GitHub stars. TypeScript · React · Transformers.js · MCP

RepoGraph Intelligence — An AI-native structural intelligence engine that analyzes codebases as dependency graphs. Extracts imports, exports, and symbols across JS, TS, and Python — calculates coupling metrics, estimates blast radius for file changes, infers architecture boundaries, and audits supply-chain dependencies against OSV.dev advisories. Includes a React Flow live graph explorer with SSE streaming and an MCP server for AI assistants. Rust core in active development. JavaScript · Rust · React Flow · Tree-sitter · MCP

Tessera — A local-first AI testing IDE that turns any codebase into a full QA dossier without sending source to the cloud. Parses the project with Tree-sitter, embeds chunks via Ollama, and generates test plans, test cases, and defect / bug reports through versioned prompts with JSON-Schema tool calling — every artifact validated against a Zod contract before review. Tauri desktop shell, AES-256-GCM-encrypted API keys, streaming previews, pluggable LLM providers (Ollama, OpenAI, OpenRouter, Anthropic). TypeScript · Rust · Tauri · React · Tree-sitter · Ollama

Strong — I build with these daily: Python, TypeScript, React, Kotlin, Flask, Git, MongoDB, Firebase

Capable — I've shipped real work using these: Rust, PyTorch, TensorFlow, scikit-learn, OpenCV, AWS, MySQL, Tailwind CSS

Exploring — Currently going deeper here: Agentic AI systems, MCP integrations, ML theory, systems-level Rust

I don't chase frameworks or trends. If something works, I want to understand why it works — not just how to call the API. I learn by building, failing, and reading papers when the stack traces stop making sense. I'm skeptical of hype and allergic to vague claims, including my own.